I've had the 5V regulator fail on two Nanos after only a few hours of operation. My circuit is for model railroad signaling and uses 9V dc for Vin. Each Nano drives two LEDs each via 1KR and powers two IR detector modules from the 5V (pin 27) each draws about 35mA. I have measured total current load for this circuit at 105mA. The Nano still works when powered from USB or 5V into pin 27 but will not work when supplied with Vin > 6V. So I conclude that the 5V regulator part has failed.
I have to confess that the Nanos are clones sourced from "the Bay"; I needed to keep costs down as my railroad signaling system uses 8 Nanos so far.
Has anyone else had this type of problem? I would think that 105mA per Nano is not a heavy load. I could run all the IR sensors from a separate 5V supply but that would involve some re-wiring that I would rather avoid.
Can you show a photo (of the voltage regulator) ? How small is it ?
(9V - 5V) * 105mA = 420 mW.
That is not something you can ignore. A small voltage regulator will get overheated.
Perhaps there are not enough decoupling capacitors, causing the voltage regulator to oscillate to make it worse.
When a voltage regulator gets too hot, it lowers the current. In most cases it should not fail. But it will get destroyed by a current peak.
You could add a few diodes in the power line to VIN to lower the voltage.
You could add a DC/DC-converter to make 5V and power the Nano via its 5V pin.
What kind of 9V is that ? Does it have peaks or really bad noise ? Sometimes cheap power supplies leak 110/230 mains voltage to the secondary side.
If the 9V is turned off, is there still 5V on the circuit somewhere ?
Hi Koepel,
Thanks for yr prompt reply. The regulator chip is AMS1117 5.0 in an SOT-223 package. I think this is the standard chip on Nanos. It appears to have decoupling capacitors. There is very little PCB area to dissipate the 420mW as you calculated above. The data sheet sates the following:
"Thermal Considerations
We have to take heat dissipation into great consideration when output current or differential voltage of input and output voltage is large. Because in such cases, the power dissipation consumed by AMS117 is very large. AMS1117 series uses SOT-223 package type and its thermal resistance is about 20°C/W. And the copper area of application board can affect the total thermal resistance. If copper area is 5cm*5cm (two sides), the resistance is about 30°C/W. So the total thermal resistance is about 20°C/W + 30°C/W. We can decrease total thermal resistance by increasing copper area in application board. When there is no good heat dissipation copper are in PCB, the total thermal resistance will be as high as 120°C/W, then the power dissipation of AMS1117 could allow on itself is less than 1W. And furthermore, AMS1117 will work at junction temperature higher than 125°C under such condition and no lifetime is guaranteed."
If I understand the thermal resistance model correctly, if this application has a TR of 120 deg C/W then my junction temp should rise by approx 50 deg but still be well under the 125 deg limit.
So in summary I think this may just be a poor quality chip that could not handle the 420mW load. I will probably convert the whole system to run from 5V applied to pin 27 on each Nano.
Cheers.
You did not answer the question about the battery type. If this is a typical 9V block, that alone explains the behavior of the system. The current capacity in mAh is very very small!
Hi Rudolf, thanks for your comment. My system has 8 Nanos all connected via "Vin" to one 9V plug-pack which is a small switch-mode supply rated at 1500mA. There is no battery. The total load on this PS should be 800~900 mA so the PS should be comfortable with that.
Only two of the 8 Nanos failed but at different times one was about 3 months ago and it was replaced and the system worked fine. The second failure was this week. It is interesting that the failure mode is identical in both Nanos; the Vin appears to be open circuit however both Nanos work correctly with 5V input either from USB or pin 27. Each Nano is mounted on a small terminal adapter PCB (purchase from Ebay); it may also be significant that the failed Nanos were each mounted in the same adapter at the time of failure. I was immediately suspicious of this but it still could be a coincidence. Of course I have conducted thorough checking of the circuit and loading connected to this adapter and found no errors.
Apart from a weak regulator (my previous reply) the only other possibility I can see is a temporary short on the 5V line for this part of the system. As noted above each Nano drives 6 LEDs that are configured as two pairs of three with common anode connected to 5V via a 1k resistor. The cathode of each LED is connected to a digital pin which will be driven LOW to turn on the LED. An over current condition from this part of the circuit is unlikely because all the ports and LEDs remain functional and the 1k resistor would limit current in any faul condition that I can envisage. Another possibility for a temporary short on the 5V is the two IR modules. These are connected to 5V, GND and their output (active low signal) is connected to an Analog input pin (A0 and A1). It seems unlikely that these units could cause a short on 5V to blow the Nano regulator and still be fully functional after that event. So ruling out the above to fault scenarios the only viable cause of a short would have to be wiring that was moved or disturbed to the point where it shorted 5V to GND somehow. I'll keep looking to see any point where this could occur but it all looks well isolated at present.
I'm about to plug in another Nano in this spot (after much cold and hot checks) so time will tell. I will let you know the outcome.
Has anyone else experienced clone Nanos with 5V regulator failures?
cheers
Yes CrossRoads, running the Nanos on 5V would have been a better system design. I did not realize how puny the heatsinking of the Nano's 5V regulator is. I will probably re-wire my system to run on a 5V supply once I dig up a suitable 5V wallwart from my odds and ends store.
cheers
This one of those times where I wish the Arduino team had just specified a 5V supply from the start, and not put in 5V regulators to accomodate old 9, 12, 15V supplies that folks might have had laying around.
Yes, it would be interesting to do some thermal measurements and determine the actual thermal resistance of the Nano 5V regulator. I read in one spec that the Nano 5V supply is rated to 200mA max load. Based on my measurements, the Nano itself is drawing about 20mA. The real issue of course is voltage drop across the regulator which will determine how much the 200mA figure needs to be de-rated. Another unknown factor in my case is quality of the regulator IC used in my clone Nanos.
I have plugged in a third Nano and the system is working OK for now. I certainly plan to change it over to all 5V supply as soon as possible.
